The Selloff and the Signal
The selloff in technology stocks this week startled some investors. It shouldn’t have. The signals of an AI bubble have been flashing for some time: billion-dollar raises for companies with no product, multibillion-dollar valuations for companies with no revenue, and nine-figure offers made to individual researchers. The astonishing part is not that the market finally paused, but that it took so long.
And then there’s the extraordinary headline number: OpenAI’s long-term spending commitments are now estimated at $1.4 trillion—roughly 1.2% of global GDP. That figure alone would have defined an entire economic era in previous decades. Today, it’s a line item in AI infrastructure planning.
Dotcoms Have Nothing on Us
A frenzy like that is enough to make you long for the relatively sane and responsible days of the Pets.com sock puppet or the synthetic CDOs. But bubbles are tricky. Are they driven by irresponsible speculators who aren’t trying to invest in great companies, but to buy something they can flip to “a greater fool?”
Another perspective is that they create a big payoff for consumers, not investors. Buyers of telecoms firms’ junk bonds in the late 1990s lost substantially, but this helped drive down bandwidth costs, supporting YouTube and Netflix. America and Britain benefit greatly from rail networks whose construction turned out very badly for the original investors.
The AI race is building products that are economic complements to one another—you need the turbines that power the grids, that power the chips, that run the models, that power the products. And you need firms to build their growth and hiring plans around the expectation that ever more of their work will be done by AI. Also, every company and every employee will be automating different sets of tasks.
If TSMC builds hugely expensive chip factories, but the big AI labs all decide they’ve spent as much as they need to, those factories are a stranded asset. But when asset prices signal that the technology is real and that the economics will be compelling, they encourage complementary investments that actually make it happen.
Something profound is happening—something we’ve seen before.
The Function of a Bubble
The popular narrative is that bubbles represent speculative greed—people buying something they hope to flip to someone more gullible. That’s a shallow reading. A more generous interpretation is that bubbles act as wealth transfers: investors overpay for infrastructure that later becomes a public good. The wreckage of dotcom excess built broadband networks. Failed railroads built the transportation grid. Overextended utilities enabled electrification.
This is true. But the deeper mechanism is more interesting.
Bubbles Create Dependent Economic Ecosystems.
A modern AI ecosystem doesn’t exist unless countless interlocking investments coincide: the turbines that power the grids, that power the datacenters, that power the GPUs, that run the frontier models, that enable the applications, that transform real industries.
Each layer is useless without the others. No single actor takes that leap unless they believe every other actor will take their leap too.
Capital moves with this coordinated co-dependency. That coordination requires belief. And belief inflates prices. Prices signal that a codependent investment and spending cycle is spiraling upwards.
Prices are a message, not just a measure.
The Capital Machine
OpenAI’s massive capex signals to everyone else—model builders, enterprise buyers, lawyers, coders, writers, startups—that a more powerful class of models is coming. Meta improves its capital expenditure plans and sends the same message. Nvidia sells through another tranche of accelerators, and the message intensifies.
Markets react to demand. They also create it.
We’ve seen this before. The auto industry scaled because car companies assumed gasoline would be cheap and abundant, and oil exploration scaled because drillers assumed cars would be everywhere. Electrification worked because utilities wired up houses before appliances existed, and GE and RCA designed appliances before utilities had finished wiring the houses. Moore’s law thrived because chipmakers built power that the software industry had no immediate use for—and the software industry wrote applications that no contemporary hardware could run efficiently.
This codependent relationship is how technological developments and technological eras emerge.
It is not efficient, cautious, or linear – and it works.
Irrational, Exuberant… and Necessary
The signs of excess are obvious and do not escape the notice of countless critics. What these critics miss is timing. Signs of a bubble do not indicate the cycle is over. They indicate the cycle has begun.
Housing Bubble? Nothing to See Here…
Warnings that the housing market would collapse circulated years before the 2008 financial crisis. An investment researcher once circulated an essay called “A Home Without Equity Is Just a Rental With Debt”, warning that house-price appreciation was driven by loosening underwriting standards and would inevitably lead to a collapse—but it was dated June 2001. Even at the post-crisis low, a decade later, the Case-Shiller index of American house prices was still 18% above its 2001 level.
Tech Exuberance Irrational?
The Netscape IPO was derided as “nutty” in 1995. A Wall Street Journal article on the Netscape IPO, published June 1995, quoted an enthusiastic buyer saying, “I don’t know anything about the company.” That was supposed to indicate a bubble about to burst. But, seven years later, at the post-dotcom low in 2002, the NASDAQ 100 was still 40% higher than in 1995.
Signs of a bubble aren’t necessarily signs that it’s time to sell. Bubbles precede the peak of the mania by an unpredictable amount. Anyone who read the (quite cogent) arguments against buying a house in 2001 or buying tech stocks in 1995 would have benefited financially from completely ignoring them.
I Believe, I Believe…
The apocryphal quote attributed to John Maynard Keynes is that markets can remain irrational longer than you can remain solvent—not because investors ignore information, but because economic decisions are more about human behavior and psychology (thank you, Daniel Kahneman). Belief creates conditions, and those conditions create further buying or selling.
Recessions end when people spend as if the recession is ending. Booms persist if enough people believe they will.
This is not irrational. It is how macro systems behave.
A Time Bubble
If you want to be early – to build, use, and profit – the signal is not subtle: it’s now. Asset prices are telling every serious participant that the future is arriving faster than the models predict.
This temporal clustering accelerates innovation. It forces capital, talent, infrastructure, and experimentation to converge simultaneously. It produces extraordinary waste. And extraordinary breakthroughs.
Inevitably, as it always does, this will overshoot what’s necessary. But, excess investment is the foundation for longer-term benefit from the creation of a substantial new market.
The Real Question
AI is in a bubble, companies will fail, and capex is unsustainably high.
The real question is whether the infrastructure being built now will unlock a technological era that outlasts the speculation that paid for it.
History suggests yes. The pattern repeats because the pattern works.
The bubble is not the danger. Missing the moment is.